This article is part of the topic “Coordination and Context in Cognitive Science,” Christo-pher T. Kello (Topic Editor). For a full listing of topic papers, see http://onlinelibrary.wiley.com/journal/10.1111/(ISSN)1756-8765/earlyview
Creating Time: Social Collaboration in MusicImprovisation
Ashley E. Walton,a Auriel Washburn,b Peter Langland-Hassan,c
Anthony Chemero,c,d Heidi Kloos,d Michael J. Richardsone
aDepartment of Statistics,Harvard UniversitybCenter for Computer Research in Music and Acoustics, Stanford University
cDepartment of Philosophy, University of CincinnatidDepartment of Psychology, University of Cincinnati
eDepartment of Psychology and Perception in Action Research Centre, Faculty of Human Sciences,Macquarie University
Received 1 September 2016; received in revised form 19 June 2017; accepted 8 September 2017
Abstract
Musical collaboration emerges from the complex interaction of environmental and informa-tional constraints, including those of the instruments and the performance context. Music improvi-sation in particular is more like everyday interaction in that dynamics emerge spontaneouslywithout a rehearsed score or script. We examined how the structure of the musical context affordsand shapes interactions between improvising musicians. Six pairs of professional piano playersimprovised with two different backing tracks while we recorded both the music produced and themovements of their heads, left arms, and right arms. The backing tracks varied in rhythmic andharmonic information, from a chord progression to a continuous drone. Differences in movementcoordination and playing behavior were evaluated using the mathematical tools of complexdynamical systems, with the aim of uncovering the multiscale dynamics that characterize musicalcollaboration. Collectively, the findings indicated that each backing track afforded the emergenceof different patterns of coordination with respect to how the musicians played together, how theymoved together, as well as their experience collaborating with each other. Additionally, listeners’experiences of the music when rating audio recordings of the improvised performances wererelated to the way the musicians coordinated both their playing behavior and their bodily move-ments. Accordingly, the study revealed how complex dynamical systems methods (namely recur-rence analysis) can capture the turn-taking dynamics that characterized both the social exchangeof the music improvisation and the sounds of collaboration more generally. The study also
demonstrated how musical improvisation provides a way of understanding how social interactionemerges from the structure of the behavioral task context.
Keywords: Music; Improvisation; Music perception; Complex systems; Social interaction;Collaboration; Recurrence analysis; Embodiment
1. Jumping in the pool
I think that he is freer than I am. I think it comes to personality too . . . I think I’mmore anxiety driven, you know . . . I feel like I’m positive, but I’m wary. I don’t knowthat he is wary. I think he’s ready to jump in the pool. And I’m like aaww oo maneehh . . . ok I’ll jump in the pool. And he’s just like . . . let’s jump.
The angst and apprehension expressed by this musician when describing his experienceimprovising brings to light the demands improvisation makes on a musician’s ability tonegotiate the uncertainty that characterizes unstructured performance contexts. It is thislack of structure that is the source of its complexity, its unpredictable nature that makesit both a challenging subject of study as well as a useful example of the under-determina-tion of most of human behaviors. Music improvisation is situated within a culture and tra-dition of musical performance whose genres, scales, chords, and motifs can provideguidance for navigating these unrehearsed exchanges. But even with these structural ele-ments, improvisation is still capable of generating wonder and surprise in both performersand listeners. Take for example the transcription of a brief excerpt of the improvisationbetween two piano players provided in Fig. 1.1 These musicians perform with an open-ended backing track, a rhythmic-less drone consisting of the pitches D and A. At thepoint where the score begins they have already been improvising for about 2 min, tradinghighly chromatic, rapidly moving lines. As Player 1 is finishing a descending line, hereaches a slowing point, and in this brief rest the improvisation takes on new life. Theplayers arrive at a sudden harmonic congruence (at measure 5) where at the samemoment Player 2 lands on the major triad associated with the tonal center, Player 1begins to outline notes of the triad mixed in with natural upper harmonic extensions. Inthis moment, the musical quality shifts significantly where the chromatic tendencies thatpreviously characterized the performance are now abandoned and the harmony remainsrooted in D major for the remainder of the performance.
In attempting to understand how these musicians are able to coordinate this shift inmusical expression, it is worth giving attention to the use of the word “land” to describehow Player 2 arrives at playing the harmonic triad. It might sound as if this harmonic
Correspondence should be sent to Ashley E. Walton, Department of Statistics, Harvard University, ScienceCenter SC 702, 1 Oxford St No. 7, Cambridge, MA 02138, USA. E-mail: [email protected]
2 A. E. Walton et al. / Topics in Cognitive Science (2017)
transition was the result of some outside physical force, which by chance happened tocoordinate with Player 1’s actions. But when considering what they are playing beforethis shift, there is nothing explicit in the musical structure that indicates this transition.And yet somehow they arrive there, and arrive there together, collectively moving thesound into a new space of musical expression. An analysis of the previous seconds, min-utes, or hours of performance, or a detailed account of their musical education and per-formance history will not provide a definitive answer about how musicians “know” whatto play—and when—in a particular performance. In the opening quote, for instance, themusician is not describing characteristics about himself and his coperformer that arespecific to musical experiences; His “wariness” and the “free” nature of his coperformershape their behavior in social contexts in general. Differences in training and performancehistory can make individuals very different musicians, but they are also very differentpeople. And in instances of musical improvisation where the indeterminacy is much likethat of everyday social interactions, what it is that makes the musicians different peoplecan significantly contribute toward the performance dynamics that emerge.
A significant proportion of contemporary research in music performance has beenfocused on understanding what musical interaction can tell us about human interaction ingeneral (D’Ausilio et al., 2015; Moran, 2014; Sawyer, 2003), specifically empathy and itsrelationship with musical production and perception (Clarke, DeNora, & Vuoskoski,2015; Greenberg, Rentfrow, & Baron-Cohen, 2015; Krueger, 2013; Rabinowitch &Knafo-Noam, 2015; Sevdalis & Keller, 2012). A few researchers have developed uniquemethods that explore social abilities within improvisatory coordination (Hart, Noy, Feni-ger-Schaal, Mayo, & Alon, 2014; Novembre, Varlet, Muawiyath, Stevens, & Keller,2015; Noy, 2014) where leader and follower roles are allowed to emerge and developspontaneously within the task constraints. Music improvisation in particular lends itself tothe examination of social connection, where in producing a cohesive performance playersmust listen and respond to each other’s musical actions. Communication between per-formers is often understood to entail the abstraction of musical production into
Fig. 1. Musical score of a selected excerpt of one of the improvised performances included in this study.
A. E. Walton et al. / Topics in Cognitive Science (2017) 3
representational imagery (Keller, 2008; Keller & Appel, 2010), but in the case of impro-visation this communication is also considered to be specified in the movement dynamicsof performers (Walton, Richardson, & Chemero, 2014). When playing without the guideof musical notation, the ongoing movements of musicians operate to construct and con-strain the flow of the performance from moment to moment (Walton, 2016).
Current research has also focused on how “embodied” musical performance is, or theextent to which body movement is a primary constituent of musical knowledge (Geeves& Sutton, 2014; Leman & Maes, 2015; Maes, 2016; Maes, Leman, Palmer, & Wanderley,2014). However, discussion on whether motor processes are “effects,” “mediators,” or“causes” does not necessarily contribute to the understanding of how individuals actuallyengage in such a skill. Take the famous anecdote of Herbie Hancock recalling a perfor-mance with Miles Davis where he thought he played a “wrong note,” and Davisresponded in such a way that “made it right”—inspiring Brian Eno’s oblique strategy:“Honor your mistake as a hidden intention.” The point here is not whether the musicalaction was a mistake or an intentional act, but that in the context of improvisation per-formers have the choice to treat it as either. Accordingly, rather than simply exploringthe self-evident fact that musical performance is embodied, one should explore how thedifferent environmental constraints and information available shapes the behavior thatemerges, where movements are part of the continuous information available to performersin guiding their musical actions. Indeed, listening and responding require the physicalmovements of each player’s musical production to continuously anticipate and adapt tothose of their coperformer’s, and so the music is in fact constituted by the self-organizeddynamics of bodily coordination that emerge from this process (Borgo, 2005; Linson &Clarke, in press; Walton, Richardson, Langland-Hassan, & Chemero, 2015). We call theempathy that enables this sort of skillful, unreflective, improvised coordination “sensori-motor empathy” (Chemero, 2016) to differentiate it from the explicit theorizing or simu-lation of the mental states of others (Gallese, 2001; Stueber, 2006). Here the act ofmusicians allowing their musical expression to be constrained by their coperformers takesthe social, communicative, and even aesthetic component of improvisation and grounds itin the body.
The experiments described in this article examine the coordination that emerges whenskillful jazz musicians “jump in the pool” together. This takes on the challenge of under-standing these spontaneous embodied musical exchanges by applying and expanding uponnonlinear methods of dynamical systems to quantify the coordination that emergesbetween improvising musicians. These methods have proven their ability to reveal newstructure within the complexity of music performance (Demos, Chaffin, & Kant, 2014;Loehr, Large, & Palmer, 2011; Maes et al., 2014) as well as pedagogy (Laroche &Kaddouch, 2015) and are particularly well suited for improvisation (Walton et al., 2015).Due to the open nature of improvised performances, it is difficult to make specific predic-tions about what exactly musicians will play and when. What these methods can reveal ishow the dynamics of musician’s musical movements adapt in response to changes in theenvironment and the actions of coperformers, and provide insight into which kinds ofdynamics result in more successful and creative performances. Accordingly, the first
4 A. E. Walton et al. / Topics in Cognitive Science (2017)
experiment applies these methods to examine the self-organized patterns of coordinationbetween the bodily movements and musical expression that emerges between pairs of jazzpianists within the context of different performance constraints. Experiment 2 looks at howthese patterns of coordination affect listener evaluations of the music being produced.
2. Freedom and constraint in music improvisation
2.1. Method
2.1.1. ParticipantsSix pairs of musicians were recruited from the local music community as well as the
University of Cincinnati’s College–Conservatory of Music (CCM). The pairs had 8–46 years of training in piano performance (M = 19.7, SD = 12.5) and 4–46 years ofexperience with improvisation (M = 14.9, SD = 13.5) and ranged in age from 18 to59 years (M = 30.4, SD = 14.3).
2.1.2. Procedure and designParticipants played standing with an Alesis Q88, 88-key semi-weighted USB/MIDI
keyboard controller (Cumberland, RI, USA), directly facing one another while theirmovements were recorded using a Latus Polhemus, wireless motion tracking system (at96 Hz; see Fig. 2). Participants were equipped with motion sensors attached to their fore-head, and both their left and right forearms (positioned on the forearm directly above thepoint where the wrist bends). Ableton Live 9.0.5 (Berlin, Germany) was used to recordall of the MIDI key presses and the resulting audio signal during the musical improvisa-tion. Pairs were instructed to develop 2-min improvised duets under vision and no-visionconditions, over different backing tracks. The vision and no-vision manipulation simplyinvolved placing a curtain between musicians for half of the performances.
Musicians improvised with two different backing tracks: a swing and a drone backingtrack.2 The swing backing track was the bass line of a chord progression from the jazzstandard used by Keller, Weber, and Engel (2011) titled: “There’s No Greater Love.”This track has a key and time signature (4/4), with the chord changes played on loop inan octave lower than that typically used. Finally, the drone backing track was a pair ofpitches, D and A, that were played for the entire duration of the 2 min. While both of thetracks reference certain genres of music, drones being essential to the harmonic base oftraditional Indian classical music, and the chord progression of the swing track highlycharacteristic of performing any jazz standard, the latter was most likely a more promi-nent part of most of the musicians’ training. Because drones are not used in jazz as often,the swing backing track (i.e., chord progression of a jazz standard) was more characteris-tic of their previous performance experiences.
At the beginning of the experiment, the musicians first performed two individualwarm-up trials, where they improvised once over each backing track while their copartici-pant sat outside the performance room. Then together the pairs performed two blocks of
A. E. Walton et al. / Topics in Cognitive Science (2017) 5
four 2-min improvised duets for a total of eight duets (2 conditions 9 2 backingtracks 9 2 blocks). After the musicians completed all of the improvisational trials, theyeach were interviewed separately. During these interviews the video and audio from thelast block of trials was played back using a laptop computer. The interview method usedwas an adaptation of that used by Norgaard (2011). The following was been read by theexperimenter before the interview:
As you are watching and listening to your performance, try to narrate your consciousthinking, considering questions like, “Where did that come from?” We are looking fora narration similar to a director’s commentary on a DVD. We are particularly inter-ested in how you are able to play with the backing track as well as with your co-per-former. We are also interested in your creative process. How are you making decisionsabout what to play and when in your improvisation?
As the musicians watched the videos of their performances, the experimenters askedfollow-up questions to explore and clarify points, as well as probed on interesting themesthat emerged in their descriptions (Charmaz, 2006).
2.1.3. AnalysesThe coordination that emerged between the improvising musicians with respect to their
body movements and their musical production was analyzed using cross-recurrence
Fig. 2. Musicians performed improvised duets while their head, left arm, and right arm movements wererecorded as well as the music produced on 88-key MIDI keyboards.
6 A. E. Walton et al. / Topics in Cognitive Science (2017)
quantification analysis (CRQA). CRQA is a nonlinear analysis method that quantifies thedynamic (time-evolving) similarity between two behaviors by identifying whether behav-ioral states in the two series reoccur over time (Marwan, 2008; Zbilut & Webber, 1992;Webber & Zbilut, 1994). Defined at a more intuitive level, CRQA assesses whether thepoints in behavioral event series visit the same states over time and then quantifies thedynamic patterns of these time-evolving recurrences using a range of different statistics.Here we report the two most commonly employed statistics: percent recurrence (%REC),which measures the percentage of recurrent points between two behavioral time or eventseries, indexing the amount of shared activity between the two behavioral systems; andMaxLine, which corresponds to the length of the longest sequence of recurrent statesobserved between two behavioral time or event series. This indirectly indexes the strengthof the localized state coupling, manifested as repeated state sequences or behavioral tra-jectories that exist between the two behavioral systems (Richardson et al., 2007). CRQAhas been used to examine the nonlinear dynamics of numerous forms of interpersonalcoordination, including rhythmic interpersonal synchrony (Richardson, Marsh, & Schmidt,2005), the postural entrainment of conversing individuals (Shockley, 2005; Shockley,Santana, & Fowler, 2003), and interpersonal timing behavior (Coey, Washburn, &Richardson, 2014), as well as the non-stationary coordination of gestural activity and ver-bal communication between adults and infants and their caregivers (e.g., Dale & Spivey,2005, 2006; Fusaroli, Razczaszek-Leonardi, & Tyl!en, 2014; Romero, Fitzpatrick, Schmidt,& Richardson, 2016). The strengths of this analysis are that it does not require any a pri-ori assumptions about the structure or stationarity of the data being analyzed, making itparticularly appropriate for quantifying the fluctuations of improvised musical production.Also, it can be applied to wide range of different types of time series, as Webber andZbilut (2005) explain: if it “wiggles in time (physiology) or space (anatomy),” recurrenceanalysis can quantify it (p. 81). And so both the dynamics of inter-musician movementcoordination and playing behavior can be evaluated with CRQA, allowing one to measurethe changes in both the musical structure (i.e., notes) as well as the interpersonal motorcoordination using the same analysis metrics.
2.1.3.1. Cross-recurrence quantification analysis of notes: MIDI output provided thenumber of each key (from 1 to 88) as well as the on and off time for each key pressed.This was used to create an event series for each player that captured what notes and/orcombinations of notes they pressed for each time point in the performance. Each keypressed was first assigned a number 1–12 representing the note of that key (A, A#, B,C#, etc.) so that a key press for a given note would be considered recurrent if the samenote was repeated, regardless of the octave. Then the total number of unique combina-tions of notes pressed by both players was determined and each was assigned a codenumber. This code number was then substituted for the note or note combination it repre-sented in order to generate a time series of code numbers for each player (Figs. 3 and 4).Timing of the playing behavior was preserved by using non-repeating random numbers torepresent points in the performance when no keys were being pressed. Because CRQA isused to examine patterns that recur within a time series, non-repeating random numbers
A. E. Walton et al. / Topics in Cognitive Science (2017) 7
Fig. 3. The MIDI output of one player from Ableton provided information about the keys musicians pressed,as well as the duration of each key pressed, at a resolution of 96 Hz. Step 1 shows how a time series wasgenerated that captured the number of each key pressed at each time point as well as when there were nokeys pressed. In Step 2, the note of each key is identified and assigned a number 1–12 such that a “C” wasconsidered the same note regardless of which octave it was played in. Step 3 demonstrates how a code num-ber was then generated for each unique combination of notes pressed. Time is preserved in the time series byinserting a non-repeating random number for time points where no key was pressed. The random numbers inPlayer 1 times series are positive, the random numbers in Player 2’s time series are negative, to assure thatthey do not count as recurrent points. Then Steps 1–3 are repeated for the MIDI output of the second player,with the same set of code numbers used to identify notes and combination of notes played. The two time ser-ies are plotted against each other to identify when the musicians repeated each other’s musical states. Thiscaptures when musicians play the exact same thing at the same time (when a yellow recurrent point falls onthe blue diagonal line), but more importantly when they repeat each other’s notes later in the performance(as when Player 2 repeats the note combination “700” after Player 1).
Fig. 4. The illustration of a portion of the recurrence plot from Fig. 3 (left) in relation to the actual recur-rence plot from the full performance (left). %REC is calculated according to the percentage of “light blue” inthe recurrence plot, MaxLine represents the length of the longest diagonal line of light blue points on theplot.
8 A. E. Walton et al. / Topics in Cognitive Science (2017)
instead of zeros assured that “non-playing” moments were not quantified as recurrentstates. While these “non-playing” moments are just as important to musical composition,this analysis focused specifically on how musicians coordinate the harmonic componentof their playing behavior.
2.1.3.2. Cross-recurrence quantification analysis of key press timing: In order to capturea more general measure of the rhythmic similarity in the timing of the musicians’ playingbehavior, an event series was created for each player that represented the timing of theirkey presses. If any key or keys were pressed at a given time point that time point wasgiven the value of “1” (for illustration, see Fig. 5). If a key was not being pressed, it wasassigned a random number so that it would not be quantified as a recurrent point asdescribed above. Therefore, recurrence in this case demonstrates more broadly when themusicians were pressing keys with the same temporal pattern, regardless of which keyswere pressed. This analysis was meant to capture the complexity in how musicians repeatspecifically the temporal aspect of each other’s playing sequences, often not repeating eachother’s exact notes but transposing riffs or repeating “shapes.” It can also capture how lefthands are engaged in “comping” or synchronizing behavior in order to create a rhythmicfoundation for the performance (note that the squares along the main diagonal indicated inblue on the cross-recurrence plot correspond to completely synchronous playing behavior).
2.1.3.3. Cross-recurrence quantification analysis of movement: Continuous CRQA wasperformed on the side-to-side movements of the musicians’ heads, left arms, and rightarms that were captured using the Polhemus wireless motion sensors.3 Continuous CRQAhas been used in the authors’ previous work to quantify coordination in interpersonaloscillatory limb movements (Richardson, Lopresti-Goodman, Mancini, Kay, & Schmidt,
Fig. 5. For key press timing the same MIDI output is used for each player, except in Step 1 any instance ofany key or combination of keys being played is given a “1” in the time series, and an instance where nothingwas played is assigned a random number. The random numbers in Player 1’s times series are positive, therandom numbers in Player 2’s time series are negative, to assure that they do not count as recurrent points.
A. E. Walton et al. / Topics in Cognitive Science (2017) 9
2008; Richardson et al., 2015). It differs from the categorical CRQA used to assess musi-cal coordination in the MIDI data in that it determines the degree of recurrent activityusing a method called phase space reconstruction. For a detailed explanation and exam-ples of the application of this method, see Webber and Zbilut (2005) and Richardsonet al. (2008).
2.2. Results
A repeated measures analysis of variance (ANOVA) was used to assess differences incoordination (i.e., in notes, key press timing, and body movements) when the musiciansimprovised with the swing versus drone backing track. Because pairs performed in eachexperimental condition twice, the coordination measures for these trials were averagedfor each of the six pairs and submitted to 2 9 2 ANOVAs that included both the backingtrack (swing/drone) and the vision manipulation (vision/no vision). Because there wereno significant main effects of vision, nor any interaction effects of vision and backingtrack, only the main effects of backing track are reported.
2.2.1. Coordination in the musicians’ playing behaviorThe results of the CRQA for coordination in the musicians’ playing behavior are
summarized in Table 1. For the combination of notes played by the musicians, therewas no significant main effect of backing track for %REC, F(1, 5) = 4.41, p = .098,g2p = .543, but there was for MaxLine, F(1, 5) = 6.61, p = .050, g2
p = .569, wheremusicians repeated each other’s note combinations in significantly longer sequenceswhen improvising with the drone backing track compared to the swing backing track.For key press timing, there was also a significant effect of backing track on %REC,F(1, 5) = 1,561, p = .011, g2
p = .757, as well as MaxLine, F(1, 5) = 14.81, p = .012,g2p = .718, where musicians repeated each other’s key press timing more and in
longer sequences when performing with the drone backing track compared to whenperforming with the swing backing track.
Table 1Results of CRQA for both the notes/combination of notes played as well as key press timing
10 A. E. Walton et al. / Topics in Cognitive Science (2017)
2.2.2. Coordination in the musicians’ movementsContinuous CRQA was used to evaluate the coordination that emerged in the side-to-
side movements of the musicians’ heads, left arms, and right arms. Again, for clarity, theresults are summarized in Table 2. For %REC of the musicians’ head movements, therewas no difference across backing tracks, F(1, 5) = 5.89, p = .060, g2
p = .541. However,MaxLine was significantly higher for the drone track, F(1, 5) = 6.56, p = .051,g2p = .568, indicating that the musicians repeated each other’s head movement sequences
more often when improvising with the drone compared to the swing track. The musi-cians’ left arm movements were also more recurrent with each other when performingwith the drone track, such that there was a significant effect of backing track for %REC,F(1, 5) = 22.90, p = .005, g2
p = .821, as well as for MaxLine, F(1, 5) = 43.59, p = .001,g2p = .897. There was also a main effect of backing track for %REC for the right arm
movements, F(1, 5) = 30.55, p = .003, g2p = .859, with more recurrence observed when
improvising with the drone backing track compared to the swing backing track. Finally,there was an effect of backing track for right arm MaxLine, F(1, 5) = 14.07, p = .013,g2p = .738, with longer sequences of recurrent activity in the movement of the musician’s
right arms when improvising with the drone backing track.
2.2.3. Surrogate analysisA surrogate analysis was performed to evaluate how much of the coordination was the
result of musicians interacting with each other in real time, as opposed to coordinationthat would be expected from two musicians performing with the same backing track. Todo so, “virtual” pairs were created where the time series of each player was matched upwith the time series of players from other musician pairs. They were matched such thateach virtual pair was made up of time series from players performing in the same block,condition, and with the same backing track. The same CRQA analysis that was run on
Table 2Results of CRQA on the musicians’ head, left arm, and right arm movements
Cross-Recurrence Measure df F p g2p M SD
Head side-to-side %REC 5 05.89 .060 .541 Drone: 1.43Swing: 0.89
0.490.77
Head side-to-side MaxLine 5 06.56 .051 .568 Drone: 405.96Swing: 310.92
A. E. Walton et al. / Topics in Cognitive Science (2017) 11
the real pairs was used to analyze these virtual pairs and the resulting measures wereaveraged. This average represented the amount of coordination to be expected betweentwo musicians’ playing behavior just by virtue of doing the same task, in the same block,under the same vision conditions, and playing with the same backing track. Note thatsuch coordination does not correspond to chance coordination per say, but rather reflectsthe degree to which the musical coordination observed between improvising musicians isdefined and constrained by a particular backing track. A 2 9 2 9 2 ANOVA with pair type(real vs. virtual) in addition to backing track and condition was used to evaluate signifi-cant differences between the average coordination found in the virtual pairs and coordina-tion between the real pairs (same values as used in the analyses above). While the effectsof backing track for the virtual pairs were consistent with those for the real pairs (morecoordination when playing with the drone), there were no significant interactions betweenpair type and backing track or condition. The results of the surrogate analysis are dis-played in Tables 3 and 4.
The results from the surrogate analysis of the playing behavior reveal that for the keypress timing there were no significant differences between real and virtual pairs for %REC, F(1, 5) = .065, p = .810, g2
p = .013, or MaxLine, F(1, 5) = .006, p = .994,g2p = .001. However, real pairs showed significantly more coordination in the notes they
played when compared with the virtual pairs for %REC, F(1, 5) = 105.7, p = .000,g2p = .955, and MaxLine, F(1, 5) = 34.02, p = .002, g2
p = .872. The key press timingmeasure, therefore, must capture coordination that is driven primarily by the structure ofthe backing track, where the coordination of musical expression is dependent upon realtime interaction between performers.
When comparing differences in movement coordination between real and virtual pairs,overall there was not a significant difference. However, there was one significant differ-ence for %REC in right arm movement, F(1, 5) = 23.27, p = .005, g2
p = .823, wherethere was also a significant interaction between pair type and backing track, F(1,5) = 19.08, p = .007, g2
p = .792. Post hoc analyses of the interaction between pair typeand backing track revealed this difference was only significant for when the musicians
Table 3Results of surrogate analysis for CRQA on the musicians’ playing.
12 A. E. Walton et al. / Topics in Cognitive Science (2017)
were performing with the drone backing track. Interestingly, the significant difference inthe movements of the right arm showed an opposite direction of effect when compared tothe musical CRQA, specifically for the less structured performance context. It is impor-tant to note that while the movements of the right and left arms are distinguishable in theanalysis of movement coordination, the MIDI data does not specify which hand playedwhich keys. In trying to understand these results, one could speculate that the musicianswould be less coordinated in their right arm movements because they were attempting toplay melodic lines that were not highly synchronized, but more complementary to theircoperformer. This may have occurred exclusively when they were improvising with thedrone backing track because unlike the swing backing track, it did not provide guidanceor precedent for when and how they should take turns playing melodic leads (i.e., “trad-ing fours” in jazz performance). Overall, clearer interpretation of these results would befacilitated by further application of these recurrence methods to a wider range of musicalperformance contexts.
2.3. Discussion
The measures used above to quantify coordination, even though taking into considera-tion the notes played by the performers, are still far from capturing all of the complexi-ties of coordination inherent to musical expression. But even in the case that there areinfinite ways musicians can respond to each other in the pitches, chords, and timbreafforded by their instrument, these responses are still situated within patterns of synchro-nization and coordination. It is this coordination, this initial commitment to cooperationthat then allows for the performers’ individual natures to interact in ways that can signifi-cantly expand the musical possibilities. The CRQA results for both the movement andmusical coordination that emerged between musicians indicate that the musicians’ playing
Table 4Results of surrogate analysis for CRQA on the musicians’ movements
Cross Recurrence Measure df F p g2p M SD
Head side-to-side %REC 1 0.774 .419 .134 Real: 1.16Virtual: 1.34
0.230.54
Head side-to-side MaxLine 1 0.334 .588 .063 Real: 358.44Virtual: 373.16
A. E. Walton et al. / Topics in Cognitive Science (2017) 13
behavior was harmonically and rhythmically more similar when improvising with thedrone compared to when they were playing with the swing backing track, reflected by theincreased coordination in both what the musicians played and how they moved together(see Fig. 6).
These results are surprising when compared to the results of the qualitative analysisof the post-session interviews. These interviews were analyzed with grounded theory(Charmaz, 2006) using the NVIVO software package in order to identify higher orderthemes in the musicians’ descriptions of their experiences improvising (for more details,see Walton, 2016). While their playing was overall more similar and more coordinatedfor the drone, they reported experiencing more freedom in the performance. One playerdiscussed the openness of the drone backing track: “The options are pretty much endlesswith this one; there are less limitations with that one note going and you can kind of justbuild whatever you want.” Another musician explained how this lack of limitation pro-vided space to more freely engage with his coperformer: “because there wasn’t a lot ofoutside information to deal with, it allowed us to really interact with each other.” Andyet another musician went further to say that this had a significant impact on the qualityof the performance: “playing with a backing track, there is no room to grow as a whole. . . it’s tough to tell a story. We did it best on the drone, where we started out with sim-ple ideas and grew in dynamics, rhythm, and added different harmonies.” The lack ofstructure provided by the drone better supported exchange between the performers andmade it possible for musicians to work together to build their own unique musical narra-tive. Thus, the cross-recurrence analysis of both the musicians’ movement and musicalcoordination provided a way for quantifying the self-imposed structure in response to thelack of constraint of the drone backing track that would not be fully captured in verbalreports of the musicians’ experience.
The results here demonstrate the challenge in understanding the complex dynamics ofinterpersonal coordination, where understanding how these dynamics emerge from andinteract with the structure of the behavioral context is key to understanding the system’sbehavior. In this case, different performance contexts required that certain relationships
Fig. 6. Example cross-recurrence plots from a pair of musicians improvising with the swing backing track(column A) and the drone backing track (column B). The top row displays the CRQA plots and measures forthe notes the musicians played, the bottom row shows the recurrence between their key press patterns. Visualinspection of the amount of recurrence in these plots (i.e., the amount light blue) demonstrate how for thedrone backing track musicians were significantly more coordinated in their musical expression.
14 A. E. Walton et al. / Topics in Cognitive Science (2017)
within the system be more constrained and synchronized, while other relationships be morecomplementary. Furthermore, the experiences of individuals engaging in this coordinationhad more to do with the way that they were able to co-create structure, and this is relevantfor considering how the environment supports the emergence of collaboration and novelbehavior. Constraints are often understood according to how they limit possibility foraction. But here we see an example of creative freedom engendered by constraint, wherethe drone track demanded a commitment to synchronization in order to maintain a commonstructure from which could emerge turn-taking dynamics that allowed the musician’s indi-viduality to flourish. This study provides an example of how performance constraints cansupport collaborative dynamics in musical exchanges that are truly social. But these resultsin particular give an account of only the musicians’ experience of the performance con-straints. Experiment 2 explores the relationship between the movement and musical coordi-nation that emerged in these different contexts of social interaction and the quality of themusic produced, or how this musical collaboration is experienced by listeners.
3. The sounds of music
3.1. Method
3.1.1. ParticipantsAn initial sample of 61 participants from the University of Cincinnati’s participant
pool were recruited to listen to recordings of the music produced by the improvisingmusicians. Listeners had an average of 3.2 years of musical experience and 1.17 yearsexperience improvising, but 46% of listeners had no musical experience at all, and 65%no experience improvising.
3.1.2. Method and procedureSurveys were programmed and administered through the Qualtrics web-based survey
platform, hosted through an account licensed to the University of Cincinnati. The surveyconsisted of four 2-min songs, one from each condition: swing–no vision, swing–vision,drone–no vision, drone–vision. Thus, there were 24 different recordings used for the lis-tener evaluations. The link to the survey was sent to participants through e-mail, and theycould fill out the survey anywhere from any device; their only instructions were to listenwith headphones and listen to the recordings in their entirety. Participants would pressplay to listen to the recording, and when it finished playing, they were able to navigate tothe next page where they responded to three different statements about the recordingusing a 4-point Likert scale. The statements were listed as follows: “The musicians in thisimprovisation are coordinated,” “This improvisation is harmonious,” and “I enjoyed lis-tening to this improvisation.” There were no instructions provided to participants regard-ing the criteria they should use in evaluating the pieces according to these terms. Asnoted above, most of the participants had little if any musical experience, let alone expe-rience with musical evaluation, so it was not necessarily expected that each participant
A. E. Walton et al. / Topics in Cognitive Science (2017) 15
would base their rating on the same qualities of the music. The goal was to capture amore general attunement to whether a musical interaction is “coordinated,” referring to amore technical component of the music, “harmonious” which implicates the aestheticqualities of the music, and “enjoyment” or personal preference for the music. Participantsanswered these questions for each of the four pieces, and at the end of the surveyanswered demographic questions that included their age, area of study, years of experi-ence with music performance, and years of experience with improvisation.
3.1.3. AnalysesThe amount of time spent on each page of the survey was recorded, and if the partici-
pant navigated to the next page without listening to the entire song, his or her data wereeliminated. The remaining scores were then averaged across participants for each track.The Pearson’s r correlation coefficient was calculated to evaluate the relationshipsbetween these averaged listener ratings of “coordination,” “harmony,” and “enjoyment,”and the CRQA measures of the coordination in the musicians’ notes, key presses, headmovements, left arm movements, and right arm movements for each recording.
3.2. Results
Table 5 displays correlations between the listener ratings and coordination of the musi-cians playing behavior, or how much the musicians repeated each other’s notes/combina-tion of notes as well as overall key press timing. There were significant correlationsbetween listener ratings and how much the musicians coordinated their key press timingwhen improvising with the drone track. Listeners rated performances with the drone asmore harmonious when the MaxLine in the musicians’ key press timing was higher, r(10) = .669, p = .017. There was also a positive correlation with key press timing, wherethe degree that listeners enjoyed the track was associated with a higher MaxLine in keypress timing, r(10) = .632, p = .028.
There was also a relationship between the listener ratings and the movement coordination thatemerged between the improvising musicians. Table 6 displays correlations between the listenerratings and head movement coordination, as measured by the %REC and MaxLine. For %REC,there was a negative correlation between how “coordinated” the listeners rated a track and thecoordination between the musician’s head movements, r(10) = !.602, p = .038, and left armmovements, r(10) = !.617, p = .033. There was also a negative correlation between the musi-cian’s left arm movement coordination and how “harmonious” the track was rated by listeners, r(10) = !.676, p = .016. For how much the listeners reported enjoying listening to the perfor-mance, there was negative correlation between “enjoyment” and the coordination of the musi-cians’ headmovements, r(10) = !.625, p = .030.
3.3. Discussion
After examining how constraints on the dynamics of musical and movement coordinationsupport collaboration between performers in Experiment 1, Experiment 2 aimed to identify
16 A. E. Walton et al. / Topics in Cognitive Science (2017)
how these dynamics might play a role in shaping the experience of listeners. With respect tohow the coordination of the musicians’ playing was related to the listener ratings, the resultsin Experiment 2 are interesting in light of the increased coordination for the drone trackfound in Experiment 1. In Experiment 1, there was more coordination in key press timingwhen the musicians were performing with the drone backing track, and in Experiment 2,how “harmonious” listeners rated the performance was significantly related to the length of
Table 5Correlations with listener ratings and %REC and MaxLine for notes played and key press timing
Survey Question (Rated on a 4-Point Likert Scale)
Swing Drone
Note(s) Key Press Note(s) Key Press
%RECThe musicians in this improvisation are coordinated r .423 .231 .243 .487
p .171 .471 .446 .109This improvisation is harmonious r .047 .275 !.167 !.070
p .886 .386 .604 .828I enjoyed listening to this improvisation r .034 !.210 .117 .332
p .916 .516 .718 .292MaxLine
The musicians in this improvisation are coordinated r .325 !.029 !.204 .501p .303 .930 .525 .097
This improvisation is harmonious r !.230 !.350 !.041 .669*p .472 .264 .900 .017
I enjoyed listening to this improvisation r !.188 !.445 !.088 .632*p .559 .147 .787 .028
Table 6Correlations with listener ratings and %REC and MaxLine in the musicians’ head, left arm, and right armmovements
Survey Question(Rated on a 4-Point Likert Scale)
Swing Drone
Head Left Arm Right Arm Head Left Arm Right Arm
%RECThe musicians in thisimprovisation are coordinated
A. E. Walton et al. / Topics in Cognitive Science (2017) 17
the recurrent sequences in the musicians’ key press timing. This suggests that listeners maybe attuned to the way the improvising musicians worked together to build the narrative ofthe performance. It also indicates that the perception of aesthetic harmoniousness is relatedto the rhythmic similarity captured by the key press timing.
With respect to the movement coordination that emerged between musicians, the nega-tive correlations between left arm coordination and listener evaluation measures allude tothe ways in which a successful improvised performance may be about how the coordina-tion between musicians reflects complementary rather than purely synchronous behavior.For example, there are positive correlations between the degree of recurrent structure ofthe musicians’ key press timing and listener ratings, but overall negative correlationsbetween ratings and the degree of coordination in the musicians’ arm movements. Thismay indicate that the co-construction of musical expression, while involving similarrhythmic qualities of playing, requires more difference in movement across the keyboardas captured by the coordination in left arm movements. This provides a demonstration ofthe subtlety required to understand coordination in the unconstrained context of everydayinteractions (Schmidt, Nie, Franco, & Richardson, 2014). Additionally, given that the lis-teners were only listening to the audio of the performance and not watching a video, it isinteresting to find significant relationships in listener experience and musicians’ headmovements. The head movement could be considered to have an indirect relationshipwith the dynamics of the music produced, but as indicated by these results may play animportant role in the resulting temporal structure of an improvisation. This is consistentwith previous work that has explored the contribution of head movement to expressivityin solo piano performance (Castellano, Mortillaro, Camurri, Volpe, & Scherer, 2008; Sud-now, 1978), and in joint musical action might play a role in how coperformers communi-cate and establish a rhythm.
In exploring auditory perception, Clarke (2005) considers how listeners discover “whatsounds are the sound of.” He makes the case that this discovery process when listening tomusic is not qualitatively different than when we perceive sounds in our everyday envi-ronment. Sounds provide information about events in our environment, often in relationto the motion of objects, ourselves, and others. Thus, we develop a perceptual attunement,where sounds specify what is moving and happening around us. This relationship betweensound and motion is also important to our experience of musical recordings. Listening tothe radio or a CD is obviously a different case of musical engagement than going to alive concert, but the sounds of the recording still specify real events: the movements ofthe performers (Clarke, 2005). Additionally, Iyer (2002, 2004), Shove and Repp (1995),Repp (1992), and Truslit (1938) have provided key initial explorations of the differentways that constraints on biological motion, whether it be of the limbs, breath, or the heartbeat, relate to both the production and perception of the spatiotemporal properties ofmusic.
The results presented here suggest that listeners respond to not only what part of theperformer is moving, and how is it moving, but also how those movements relate to thoseof coperformers. That is, the events specified by sounds necessarily include not onlymovements of performers but also how performers are coordinating their movements
18 A. E. Walton et al. / Topics in Cognitive Science (2017)
together. In fact, examining music in relation to the action capabilities of individual bod-ies leaves out the majority of musical listening that involves hearing the coordination ofmany. When listening to a quartet, band, or orchestra, how the musicians collectivelycoordinate their musical movements is arguably the primary determinant of the quality ofthe musical experience. In the case of this study, it may be incorrect to say listeners“hear” the way the musicians are moving their heads. However, the results do seem toindicate that listeners are attuned to the way the head movements structure the dynamicsthat emerge and evolve across the time span of the performance.
The fact that the musicians’ movements affected the experience of the music is alsointeresting in light of the question of who was listening. The listeners in this study wereprimarily non-expert musicians, with virtually no experience with music improvisation inparticular. Lack of common musical education, previous experience, difference in percep-tion of criteria, language, personal preference, and in this case even the listening environ-ment are some of the many factors that introduce variability to how the listeners respondto a given piece of music at a given time. But while the importance of head movementcoordination in this study will not always prove relevant to all listening experiences, itseems possible to identify how in different contexts the self-organized interactions of asystem’s components give rise to sounds that have certain meaning to listeners (“coordi-nation,” “enjoyment,” etc.).
When considering the importance of coordination to many of our experiences, it in factis not surprising that interpersonal coordination would be an important part of our percep-tual sensitivity to sounds. Phillips-Silver and Keller (2012) argue that our ability toentrain with others and our environment begins at the very early stages of development,where at the “roots” of our ability to coordinate musically are more general turn-takingskills that involve anticipating and adapting to movements of those around us. Clarke(2005) claims that music has the power to induce the experience of musical “agents.”The results of the current study provide initial evidence for how the movement of thoseagents constitutes a common ground for our perception of music, where not only ourengagement in musical production but also musical listening is rooted in our ability tocoordinate with the world and others.
4. Making waves
We have seen, in Experiment 1, that skilled musicians playing with an unstructuredbacking track coordinate with one another more than when they are playing along with astructured backing track. From the post-session interviews, we also found that the perfor-mance context that brought about the increase in coordination created an increased oppor-tunity for interpersonal connection and opportunity for creativity. We also found, inExperiment 2, that listeners can hear the patterns of coordination that musicians createtogether and that it influences their perceptions of aesthetic features of the recordings.Taken together then, the current results reveal how different constraints on improvisationshape the experience of musicians as well as the musical and movement dynamics that
A. E. Walton et al. / Topics in Cognitive Science (2017) 19
emerge (and are heard). Discovering performance contexts capable of successfully facili-tating these dynamics is not about imposing structure on a performance, but how the con-text demands participants to codetermine that structure.
When describing his experience improvising with the drone backing track, one musi-cian explains: “There’s no time, it’s just a steady tone, . . . we created time between us.”While a lot of research aims to understand the power in our ability to synchronize, orkeep time together, when it comes to understanding music performance as a social inter-action, it is more important how we create time together. Talking about how performers’musical actions “overlap,” how they “fill in gaps” or “compensate” for one another is stillconsidering them first as individual entities who then must find a way to coexist throughthe coordination of their playing. But when musicians create time together, they are nolonger behaving as individuals but as a single, collective unit—flexibly navigating theshared temporal landscape conceived by their musical actions (Laroche, Berardi, &Brangier, 2014). Doing so requires sensorimotor empathy, the ability to skillfully andimplicitly coordinate your actions with another person. We do this in everyday conversa-tion (Dale, Fusaroli, Duran, & Richardson, 2013; Fusaroli et al., 2014; Richardson, Dale,& Marsh, 2014; Schmidt, Morr, Fitzpatrick, & Richardson, 2012; Schmidt et al., 2014;Shockley, Richardson, & Dale, 2009), but it is more visible and more striking in a musi-cal context. Collective musical improvisation demands openness and adaptation to main-tain “group flow” (Hart & Di Blasi, 2013; Sawyer, 2006), and also an immense trust inthe collective ability to build a foundation from which the performance will be devel-oped, despite the possible wariness or anxiety that this may induce. But “mistakes” or“surprises” are defined by how their actions deviate in relationship to this foundation per-formers create together. And in the concession, adjustment, and opposition to this code-termined “time” is where each performer’s individuality, their different personalitiescreate the possibility for something truly novel and unexpected.
The creation of time is about the creation of the beat, and the beat is felt physicallywith every key press, foot tap, breath, and head sway. Musicians’ movements define thestructure and can also stretch, obscure or accentuate its parts, the minor but powerfultemporal deviations from the force of the rhythm are considered to be the source of“groove” in music (Iyer, 2002; Janata, Tomic, & Haberman, 2012; Pressing, 2002). Con-sider the positive correlation between the coordination of the musicians’ key press timingand how “harmonious” the performances were rated, as well as the negative correlationbetween head and left arm coordination and the “harmonious” ratings. It seems the physi-cal rigor of creating this structure, and the ways musicians engage in the freedom to devi-ate from it, is salient to listeners. And while an expert examination of a musical scorecan provide information about harmonic congruence as demonstrated in the excerptdescribed in the introduction, what may be more universally relevant is how they negoti-ate the risks of taking these liberties. Whether it is characterized by excitement or hesita-tion, it is the manner in which the musicians “jump in the pool” that may be mostessential to the meaning of music for listeners.
It might be difficult to conceive of how this force can be felt by listeners who are notactively engaged, or even present for the improvisational performance. Studies have
20 A. E. Walton et al. / Topics in Cognitive Science (2017)
looked at how music induces movement in listeners (Hurley, Martens, & Janata, 2014;Ross, Warlaumont, Abney, Rigoli, & Balasubramaniam, 2016; Stupacher, Hove, &Janata, 2016; Van Dyck et al., 2013), but they are not bi-directionally coupled with per-formers and so cannot alter the evolution of the musical events. As discussed previously,however, their perception is rooted in their general ability to take part in this activeengagement with their environment. And so when listening to the way that musicians takerisks, adapt to, and defy the beat, their perceptual sensitivity to the musical coordinationallows them to follow alongside the performers, anticipating their musical choices(Huron, 2006) and experiencing anxiety and surprise as they succeed or fall flat.
The “sudden” harmonic congruence or surprising “mistakes” that result from thesedeviations and risks are a crucial part of how performers discover new musical spaces.As improvising musicians engage in sensorimotor empathy to co-create these spaces by“feeling-into” and “feeling-out” possibilities with their musical movements, this allowsfor the creation of new social spaces. Think about how music can “move” us. The drivingforce of a drum line in a military march or the riots following Stravinsky’s Rite of Springare some more literal examples that come to mind. But music is also capable of expand-ing our openness to unfamiliar social realms that initially make us cautious or uneasy. Itcan cultivate a willingness to inhabit these social spaces, if even for a brief moment ofmusical time, our will steadied by the universal familiarity of the rhythms of humanmotion. It is this common ground of the embodied aesthetic of musical collaboration thatinspires the motion necessary to create new possibilities, and bigger waves.
Acknowledgments
The research presented was completed as part of the first author’s master’s work (Wal-ton, 2016). The research was also supported, in part, by funds from the National Institutesof Health (R01GM105045). The authors would like thank Dr. Michael Riley (University ofCincinnati), Dr. Charles Coey (College of the Holy Cross), and Dr. Richard C. Schmidt(College of the Holy Cross) for their support and for the helpful comments they providedin relation to this work. A special thanks to Adam Petersen and Josh Jessen for assistancein conceptualizing the study and composing the backing tracks, Ben Sloan for sound engi-neering, and Stephen Patota for transcribing the included excerpt.
Notes
1. The video for this excerpt can be viewed at http://www.emadynamics.org/music-improvisation-video.
2. The musicians also improvised with an ostinato track in 7/8 time, but these resultsare omitted for this publication.
3. The phase space embedding time lag was 25 samples and 5 embedding dimensions,and a recurrence radius of 10% of the mean rescaled distance between points.
A. E. Walton et al. / Topics in Cognitive Science (2017) 21
References
Borgo, D. (2005). Sync or swarm: Improvising music in a complex age. New York: The ContinuumInternational Publishing Group.
Castellano, G., Mortillaro, M., Camurri, A., Volpe, G., & Scherer, K. (2008). Automated analysis of bodymovement in emotionally expressive piano performances. Music Perception: An Interdisciplinary Journal,26(2), 103–119.
Charmaz, K. (2006). Constructing grounded theory: A practical guide through qualitative analysis. ThousandOaks, CA: Sage.
Chemero, A. (2016). Sensorimotor empathy. Journal of Consciousness Studies, 23(5–6), 138–152.Clarke, E. F. (2005). Ways of listening: An ecological approach to the perception of musical meaning.
Oxford, UK: Oxford University Press.Clarke, E., DeNora, T., & Vuoskoski, J. (2015). Music, empathy and cultural understanding. Physics of Life
Reviews, 15, 61–88.Coey, C. A., Washburn, A., & Richardson, M. J. (2014). Recurrence quantification as an analysis of temporal
coordination with complex signals. In N. Marwan, M. Riley, A. Giuliani, & C. L. Webber (Eds.),Translational recurrences: From mathematical theory to real-world applications (pp. 173–186). Cham,Switzerland: Springer.
D’Ausilio, A., Novembre, G., Fadiga, L., & Keller, P. E. (2015). What can music tell us about socialinteraction? Trends in cognitive sciences, 19(3), 111–114.
Dale, R., Fusaroli, R., Duran, N., & Richardson, D. C. (2013). The self-organization of human interaction.Psychology of Learning and Motivation, 59, 43–95.
Dale, R., & Spivey, M. J. (2005). Categorical recurrence analysis of child language 27th annual meeting ofthe cognitive science society (pp. 530–535). Mahwah, NJ: Lawrence Erlbaum.
Dale, R., & Spivey, M. J. (2006). Unraveling the dyad: Using recurrence analysis to explore patterns of syntacticcoordination between children and caregivers in conversation. Language Learning, 56(3), 391–430.
Demos, A. P., Chaffin, R., & Kant, V. (2014). Toward a dynamics theory of body movement in musicalperformance. Frontiers in Psychology, 5(477), 1–6. https://doi.org/0.3389/fpsyg.2014.00477.
Fusaroli, R., Razczaszek-Leonardi, J., & Tyl!en, K. (2014). Dialog as interpersonal synergy. New Ideas inPsychology, 32, 147–157.
Gallese, V. (2001). The “shared manifold” hypothesis. From mirror neurons to empathy. Journal ofConsciousness Studies, 8(5–6), 33–50.
Geeves, A., & Sutton, J. (2014). Embodied cognition, perception, and performance in music. EmpiricalMusicology Review, 9, 248–253.
Greenberg, D. M., Rentfrow, P. J., & Baron-Cohen, S. (2015). Can music increase empathy? Interpretingmusical experience through the empathizing–systemizing (ES) theory: Implications for autism. EmpiricalMusicology Review, 10(1–2), 80–95.
Hart, E., & Di Blasi, Z. (2013). Combined flow in musical jam sessions: A pilot qualitative study.Psychology of Music, 43(2), 275–290.
Hart, Y., Noy, L., Feniger-Schaal, R., Mayo, A. E., & Alon, U. (2014). Individuality and togetherness injoint improvised motion. PLoS ONE, 9(2), e87213.
Hurley, B. K., Martens, P. A., & Janata, P. (2014). Spontaneous sensorimotor coupling with multipart music.Journal of Experimental Psychology: Human Perception and Performance, 40(4), 1679.
Huron, D. B. (2006). Sweet anticipation: Music and the psychology of expectation. Cambridge, MA: MITpress.
Iyer, V. (2002). Embodied mind, situated cognition, and expressive microtiming in African-American music.Music Perception: An Interdisciplinary Journal, 19, 387–414.
Iyer, V. (2004). Exploding the narrative in jazz improvisation. Uptown conversations: The new jazz studies(pp. 393–403). New York: Columbia University Press.
22 A. E. Walton et al. / Topics in Cognitive Science (2017)
Janata, P., Tomic, S. T., & Haberman, J. M. (2012). Sensorimotor coupling in music and the psychology ofthe groove. Journal of Experimental Psychology: General, 141(1), 54.
Keller, P. E. (2008). Joint action in music performance. In F. Morganti, A. Carassa, & G. Riva (Eds.),Enacting intersubjectivity: A cognitive and social perspective to the study of interactions (pp. 205–221).Amsterdam: IOS Press.
Keller, P. E., & Appel, M. (2010). Individual differences, auditory imagery, and the coordination of bodymovements and sounds in musical ensembles. Music Perception, 28, 27–46.
Keller, P. E., Weber, A., & Engel, A. (2011). Practice makes too perfect: Fluctuations in loudness indicatespontaneity in musical improvisation. Music Perception, 29, 107–112.
Krueger, J. (2013). Empathy, enaction, and shared musical experience. In T. Cochrane, B. Fantini, & K. R.Scherer (Eds.) The emotional power of music: Multidisciplinary perspectives on musical expression,arousal, and social control (pp. 177–196). Oxford, UK: Oxford University Press.
Laroche, J., Berardi, A. M., & Brangier, E. (2014). Embodiment of intersubjective time: Relational dynamics asattractors in the temporal coordination of interpersonal behaviors and experiences. Frontiers in Psychology,5, 1180.
Laroche, J., & Kaddouch, I. (2015). Spontaneous preferences and core tastes: Embodied musical personalityand dynamics of interaction in a pedagogical method of improvisation. Frontiers in Psychology, 6, 522.
Leman, M., & Maes, P.-J. (2015). The role of embodiment in the perception of music. Empirical MusicologyReview, 9(3-4), 236–246.
Linson, A., & Clarke, E. F. (in press). Distributed cognition, ecological theory, and group improvisation. InE. Clarke, & M. Doffman (Eds.), Distributed creativity: Improvisation and collaboration in contemporarymusic. Oxford, UK : Oxford University Press.
Loehr, J., Large, E. W., & Palmer, C. (2011). Temporal coordination in music performance: Adaptation totempo change. Journal of Experimental Psychology: Human Perception and Performance, 37 (4), 1292–1309.
Maes, P. J. (2016). Sensorimotor grounding of musical embodiment and the role of prediction: A review.Frontiers in Psychology, 7, 308.
Maes, P. J., Leman, M., Palmer, C., & Wanderley, M. (2014). Action-based effects on music perception.Frontiers in Psychology, 4, 1008.
Marwan, N. (2008). A historical review of recurrence plots. European Physical Journal, 164, 3–12.Moran, N. (2014). Social implications arise in embodied music cognition research which can counter
musicological “individualism.” Frontiers in Psychology, 5, 676.Norgaard, M. (2011). Descriptions of improvisational thinking by artist-level jazz musicians. Journal of
Research in Music Education, 59, 109–127. https://doi.org/10.1177/0022429411405669.Novembre, G., Varlet, M., Muawiyath, S., Stevens, C. J., & Keller, P. E. (2015). The E-music box: An
empirical method for exploring the universal capacity for musical production and for social interactionthrough music. Royal Society Open Science, 2(11), 150286.
Noy, L. (2014). The mirror game: A natural science study of togetherness. In A. Citron, S. Aronson-Lehavi,& D. Zerbib (Eds.), Performance studies in motion (pp. 318–327). London: Bloomsbury Methuen Drama.
Phillips-Silver, J., & Keller, P. E. (2012). Searching for roots of entrainment and joint action in early musicalinteractions. Frontiers in Human Neuroscience, 6, 26.
Pressing, J. (2002). Black Atlantic rhythm: Its computational and transcultural foundations. Music Perception,19, 285–310.
Rabinowitch, T. C., & Knafo-Noam, A. (2015). Synchronous rhythmic interaction enhances children’sperceived similarity and closeness towards each other. PLoS ONE, 10(4), e0120878.
Repp, B. H. (1992). Music as motion: a synopsis of Alexander Truslit’s (1938) “Gestaltung und Bewegung inder Musik”. Haskins Laboratories Status Report on Speech Research (pp. 265–278).
Richardson, M. J., Dale, R., & Marsh, K. L. (2014). Complex dynamical systems in social and personalitypsychology: Theory, modeling and analysis. In H. T. Reis & C. M. Judd (Eds.), Handbook of researchmethods in social and personality psychology (2nd ed.). New York: Cambridge University Press.
A. E. Walton et al. / Topics in Cognitive Science (2017) 23
Richardson, M. J., Harrison, S. J., Kallen, R. W., Walton, A., Eiler, B., & Schmidt, R. C. (2015). Self-organized complementary coordination: Dynamics of an interpersonal collision-avoidance task. Journal ofExperimental Psychology: Human Perception and Performance, 41, 665–679.
Richardson, M. J., Lopresti-Goodman, S., Mancini, M., Kay, B. A., & Schmidt, R. C. (2008). Comparing theattractor strength of intra- and interpersonal interlimb coordination using cross recurrence analysis.Neuroscience Letters, 438, 340–345.
Richardson, M. J., Marsh, K. L., & Schmidt, R. C. (2005). Effects of visual and verbal information onunintentional interpersonal coordination. Journal of Experimental Psychology: Human Perception andPerformance, 31, 62–79.
Richardson, M. J., Schmidt, R. C., & Kay, B. A. (2007). Distinguishing the Noise and Attractor Strength ofCoordinated Limb Movements Using Recurrence Analysis. Biological Cybernetics, 96, 59–78.
Romero, V., Fitzpatrick, P., Schmidt, R. C., & Richardson, M. J., (2016). Using cross-recurrencequantification analysis to understand social motor coordination in children with autism spectrum disorder.C. L. Webber Jr. C. Ioana, & N. Marwan (Eds.), Recurrence plots and their quantifications: Expandinghorizons, Springer Proceedings in Physics 180. Cham, Switzerland: Springer.
Ross, J. M., Warlaumont, A. S., Abney, D. H., Rigoli, L. M., & Balasubramaniam, R. (2016). Influence ofmusical groove on postural sway. Journal of Experimental Psychology: Human Perception andPerformance, 42, 308–319.
Sawyer, R. K. (2003). Group creativity: Music, theatre, collaboration. Mahwah, NJ: Lawrence ErlbaumAssociates Inc.
Sawyer, R. K. (2006). Group-creativity: Musical performance and collaboration. Psychology of Music, 34,148–165.
Schmidt, R. C., Morr, S., Fitzpatrick, P., & Richardson, M. J. (2012). Measuring the dynamics ofinteractional synchrony. Journal of Nonverbal Behavior, 36, 263–279.
Schmidt, R. C., Nie, L., Franco, A., & Richardson, M. J. (2014). Bodily synchronization underlying joketelling. Frontiers in Human Neuroscience, 8, 633. https://doi.org/10.3389/fnhum.2014.00633.
Sevdalis, V., & Keller, P. E. (2012). Perceiving bodies in motion: Expression intensity, empathy, andexperience. Experimental Brain Research, 222, 447–453.
Shockley, K. (2005). Cross recurrence quantification of interpersonal postural activity. Tutorials incontemporary nonlinear methods for the behavioral sciences, 142–177.
Shockley, K., Richardson, D. C., & Dale, R. (2009). Conversation and coordinative structures. Topics inCognitive Science, 1(2), 305–319.
Shockley, K., Santana, M. V., & Fowler, C. A. (2003). Mutual interpersonal postural constraints are involvedin cooperative conversation. Journal of Experimental Psychology: Human Perception and Performance,29, 326–332.
Shove, P., & Repp, B. H. (1995). Musical motion and performance: Theoretical and empirical perspectives.In J. Rink (Ed.), The practice of performance (pp. 55–82). Cambridge, MA: Cambridge University Press.
Stueber, K. R. (2006). Rediscovering empathy: Agency, folk psychology, and the human sciences. Cambridge,MA: MIT Press.
Stupacher, J., Hove, M. J., & Janata, P. (2016). Audio features underlying perceived groove and sensorimotorsynchronization in music. Music Perception, 33(5), 571–589.
Sudnow, D. (1978). Ways of the hand. Cambridge, MA: Harvard University Press.Truslit, A. (1938). Gestaltung und Bewegung in der Musik. Berlin: Chr. Friedrich Vieweg.Van Dyck, E., Moelants, D., Demey, M., Deweppe, A., Coussement, P., & Leman, M. (2013). The impact of
the bass drum on human dance movement. Music Perception, 30, 349–359.Walton, A. (2016). Music improvisation: Spatiotemporal patterns of coordination. (Master’s thesis). Available
from OhioLink Electronic Theses and Dissertation Center. (http://rave.ohiolink.edu/etdc/view?acc_num=ucin1457619683)
Walton, A., Richardson, M. J., & Chemero, A. (2014). Self-organization and semiosis in jazz improvisation.International Journal on Signs and Semiotics Systems, 3, 12–25.
24 A. E. Walton et al. / Topics in Cognitive Science (2017)
Walton, A., Richardson, M. J., Langland-Hassan, P., & Chemero, A. (2015). Improvisation and the self-organization of multiple musical bodies. Frontiers in Psychology: Theoretical and PhilosophicalPsychology, 6, 313.
Webber, C. L., & Zbilut, J. P. (1994). Dynamical assessment of physiological systems and states usingrecurrence plot strategies. Journal of Applied Physiology, 76(2), 965–973.
Webber, Jr. C. L., & Zbilut, J. P. (2005). Recurrence quantification analysis of nonlinear dynamical systems.In M. A. Riley & G. C. Van Orden (Eds.), Contemporary nonlinear methods for behavioral scientists: Awebbook tutorial (pp. 26–94).
Zbilut, J. P., & Webber, C. L. (1992). Embeddings and delays as derived from quantification of recurrenceplots. Physics Letters A, 171(3–4), 199–203.
A. E. Walton et al. / Topics in Cognitive Science (2017) 25
